Control of transcription initiation is the most common mechanism by which genes are regulated. An essential part of this control is the sequence-specific binding of proteins (transcription factors) to DNA sequence elements. However, there is not a simple correspondence in eukaryotes between the existence of a transcription factor binding site and the regulation of a nearby gene by that factor. The availability of genomic sequences and the advent of DNA microarray technologies allow us to ask fundamentally new questions about gene regulation. Where in the past we have been able to show that a sequence motif is necessary for regulation of a gene, or even sufficient in some context, we can now ask how predictive of regulation the motif is, evaluated over every gene in the genome. The improvement in correlation between sequence elements and expression can be evaluated as more information is added to the analysis: improved binding site descriptions, considerations of spacing and orientation, cooperative and competing interactions, and so on. Ultimately, we would like to be able to explain, on the basis of genomic sequence, why one set of genes is regulated in response to a signal and the rest of the genes in the genome are not. As a first step in this direction, we will determine the limits to how well simple binding site considerations can rationalize gene regulation and will determine the role of protein-protein competition in gene regulation by related transcription factors. These studies are being conducted in yeast because the delineation of genes in the genome is better for yeast than for higher eukaryotes and because DNA microarray data can readily be obtained for essentially every gene. In addition to computational analyses on a variety of systems, experimental data will be obtained for selected systems with properties ideally suited for addressing issues of binding specificity and competition.